How to Start a Prediction Market: The Go-to-Market Half

Search “how to start a prediction market” and you’ll find legal explainers, trading guides, and Reddit threads about how to make money betting on outcomes.

What you won’t find is much about the part that decides whether anyone actually uses the thing you built. That’s the go-to-market half, and it’s the half I know.

I launched Hard Rock Bet, Florida’s first legal online sportsbook, and took it across six states against DraftKings and FanDuel. Before that, I ran integrated marketing at Twitch, where we built prediction extensions into creator co-streams of Thursday Night Football. Prediction markets sit right where those two worlds meet.

Quick note before we start: this is a marketing playbook. The legal side of prediction markets is moving fast, and you need specialized counsel for it.

The Two Numbers That Define the Opportunity

Here’s the category in two data points.

The spike. During the first half of the 2026 World Cup, Kalshi and Polymarket took 73.5% of new installs among the major sportsbook and prediction market apps tracked by Apptopia. Kalshi alone took 42.3% (DeFi Rate).

The drop. Kalshi’s weekly sports volume peaked near $9 billion in the week ending July 5. By the week ending July 19, the week of the final, it was down to about $4 billion. Polymarket fell nearly 70% over the same stretch (The Block).

Real talk: that’s the entire go-to-market problem. Spikes are easy when the world is watching. Habits are hard. A new prediction market that plans for the spike and ignores the day after is building a seasonal business by accident.

Step 1: Decide Who You’re For

Prediction markets can cover anything: sports, politics, the economy, culture, weather. That’s the appeal, and it’s also the trap.

Sports made up about 80% of volume on both Kalshi and Polymarket during the World Cup, per The Block. If you launch as “trade on anything,” you’re competing with two giants on their home field.

Pick an audience you can own. A sport, a community, a type of question, a region. Then write one sentence explaining why someone in that audience should use you instead of the app already on their phone. If you can’t write that sentence, no amount of ad spend will fix it.

Step 2: Launch Around Moments, and Plan for the Day After

Prediction markets are event-native. Attention arrives with the big game, the election, the awards show, the earnings call.

So launch into those moments. At Hard Rock Bet, every state launch was a coordinated burst of brand, media, and cultural activations stacked into a compressed window, because that’s when attention and intent peaked.

But build the calendar past the moment. For every tentpole event, ask what the next three reasons to come back are. The weekly games after the opener. The next round of a tournament. The recurring questions your community cares about every week.

Step 3: Build the Habit in the First Week

This is where prediction markets win or lose, and it’s the same lesson I learned at Pandora and again at Hard Rock Bet.

At Pandora, we discovered that two early actions, creating more than one station and giving more than five thumbs in the first week, predicted whether a listener would stay for years. We built the whole lifecycle program around them.

At Hard Rock Bet, the equivalent behaviors were a second bet within 48 hours, betting on more than one sport, and live in-play wagering. We built campaigns to drive each one, and ARPU came in 17% above target.

Find your version. For a prediction market, the candidates look like:

• A second trade within 48 hours of the first
• Following a market so the product has a reason to bring the user back
• Trading in a second category beyond the event that brought them in
• Coming back for a recurring market that resolves weekly

Then point every email, push notification, and in-app message at those behaviors.

Step 4: Make Prediction Part of the Entertainment

The best proof I have that prediction mechanics drive engagement came from Twitch.

When we pioneered creator co-streaming for Thursday Night Football, we integrated Twitch-native features into the broadcast, including prediction extensions tied to gameplay. Audience interaction rates hit 70 to 74%. The program ran 21 creators across 200+ streams and drove 3.3 million hours watched in a single season (case study).

Fans wanted to watch with their people and have a stake in the moment. A prediction market that feels like part of the watching experience will retain better than one that feels like a trading terminal.

Step 5: Run Creators Like Your Reputation Depends on It

Creators are the fastest channel in this category. They’re also where it has gone most publicly wrong.

In June 2026, The Wall Street Journal reported that Polymarket paid creators to film staged bets. The Journal reviewed 1,105 videos from 10 creators. Across 118 of them, creators celebrated about $900,000 in winnings that never happened, some filmed on a lookalike site. The clips drew more than 140 million views (The Block).

Creator marketing works when the audience trusts the creator and the experience is real. At Twitch Sells Out, 29 creators sold more than 100 products live and drove $24 million in Prime Day sales, because creators featured products they actually used (case study).

Your creator rules should cover:

• Disclosure on every paid post. No exceptions.
• Real product, real accounts, real outcomes. No staged wins.
• Age and audience fit. Know who each creator’s audience actually is.
• A review process before anything goes live.

Step 6: Write Marketing That Holds Up

The rules are in motion. The CFTC’s interim rule excluding state-licensed gambling from the federal definition of “swap” took effect October 19, 2026, and comments on it and a companion proposal on event contracts are due November 18 (RG.org). States and courts are still fighting over who regulates what.

When Quartz asked me about prediction markets earlier this year, my point was that the same engagement loops, social validation, and low-friction onboarding that make digital platforms sticky can make wagering feel like informed opinion instead of gambling (Quartz coverage). Regulators are watching exactly that line.

So assume every ad, every creator post, and every onboarding screen will be read by a regulator and a reporter. Be precise about what the product is. Build in age verification and responsible gaming from day one.

Step 7: Know Who You’re Really Competing With

You’re competing with Kalshi and Polymarket. You’re also competing with the sportsbooks, because DraftKings and FanDuel now run standalone prediction apps alongside their sportsbooks (DeFi Rate).

The incumbents have brand awareness and budgets you can’t match. What a new market can have is focus: a specific audience, a sharper product for them, and a retention system that keeps them after the spike. We were the challenger at Hard Rock Bet too, up against DraftKings and FanDuel. Concentrated launch campaigns beat our acquisition forecasts by 232%, and the engagement programs behind them pushed ARPU 17% above target.

A 90-Day Go-to-Market Plan for a New Prediction Market

Takeaways

• Building the exchange is half the job. Go-to-market decides whether anyone comes back.
• Pick an audience you can own instead of trying to out-Kalshi Kalshi.
• Launch into moments, then plan the next three reasons to return.
• Find your activation behaviors and build the first week around them.
• Run creators and ads like a regulator is reading, because one is.

For more on how this works in practice, see the Hard Rock Bet case study, the sportsbook state launch playbook, and how Field Vision works with gaming and prediction market startups.

Frequently Asked Questions

How do you start a prediction market?

There are two halves. The first is building a compliant exchange, which requires specialized legal counsel because the rules are changing. The second is go-to-market: choosing a specific audience, launching around real-world events, building first-week habits so users return after the spike, and running creator and ad programs that hold up to regulatory and press scrutiny.

How do prediction markets acquire users?

Mostly around big events. During the first half of the 2026 World Cup, Kalshi and Polymarket took 73.5% of new installs among major betting and prediction apps tracked by Apptopia. Creators and event-driven media drive the spike. Retention programs decide whether those users stay.

How do prediction markets keep users after a big event?

By building habits early. According to David Hampian, who launched Hard Rock Bet across six states, the key is identifying the early behaviors that predict long-term value, such as a second trade within 48 hours or trading in a second category, and building lifecycle messaging around them. Recurring markets give users a weekly reason to come back.

Can prediction markets use influencer marketing?

Yes, with strict rules. A June 2026 Wall Street Journal investigation found Polymarket creators had posted videos celebrating about $900,000 in winnings that never happened. Every paid post should be disclosed, show the real product and real outcomes, fit the creator’s actual audience age, and pass a review before publishing.

How should a new prediction market compete with Kalshi and Polymarket?

With focus. Pick an audience or category you can own, give that audience a sharper product and a clear reason to switch, and win on retention. Incumbents, including sportsbooks that now run their own prediction apps, have more awareness and budget, so a new entrant wins by being more specific.

Building a prediction market? Book a strategy call and we’ll work through your audience, your launch moment, and your first 90 days. As a fractional CMO, I help gaming and prediction market startups build go-to-market plans that hold up after the spike.